Pathogenic Roles for RNASET2 in Clear Cell Renal Cell Carcinoma

Abstract

A specific splicing isoform of RNASET2 is associated with worse oncologic outcomes in clear cell renal cell carcinoma (ccRCC). However, the interplay between wild-type RNASET2 and its splice variant and how this might contribute to the pathogenesis of ccRCC remains poorly understood. We sought to better understand the relationship of RNASET2 in the pathogenesis of ccRCC and the interplay with a pathogenic splicing isoform (RNASET2-SV) and the tumor immune microenvironment. Using data from The Cancer Genome Atlas and Clinical Proteomic Tumor Analysis Consortium, we correlated clinical variables to RNASET2 expression and the presence of a specific RNASET2-SV. Immunohistochemical staining with matched RNA sequencing of ccRCC patients was then utilized to understand the spatial relationships of RNASET2 with immune cells. Finally, in vitro studies were performed to demonstrate the oncogenic role of RNASET2 and highlight its potential mechanisms. RNASET2 gene expression is associated with higher grade tumors and worse overall survival in The Cancer Genome Atlas cohort. The presence of the RNASET2-SV was associated with increased expression of the wild-type RNASET2 protein and epigenetic modifications of the gene. Immunohistochemical staining revealed increased intracellular accumulation of RNASET2 in patients with increased RNA expression of RNASET2-SV. In vitro experiments reveal that this accumulation results in increased cell proliferation, potentially from altered metabolic pathways. RNASET2 exhibits a tumor-promoting role in the pathogenesis of ccRCC that is increased in the presence of a specific RNASET2-SV and associated with changes in the cellular localization of the protein.

Introduction

Renal cell carcinoma (RCC) is among the top 10 most common neoplasms in the United States.¹ Clear cell RCC (ccRCC) is the most common histologic subtype, accounting for approximately 75% of cases.² Data from the National Cancer Institute Surveillance, Epidemiology, and End Results Cancer Statistics Review show that patients with RCC initially present with regional spread (ie, lymph nodes) or metastatic disease in 17% and 16% of cases, respectively.³ The prognosis for these patients who develop locoregional spread or metastatic disease is poor.^4,5 Our understanding of mutational drivers and tumor immunology in ccRCC has led to the development of targeted agents and immunotherapy, yet long-term survival remains poor.⁶ Additionally, critical questions remain about what drives pathogenesis and recurrence among localized ccRCC patients.

In contrast to research on the mutational burden in ccRCC, the study of alternative and aberrant mRNA splicing as potential biomarkers, drivers, and therapeutic targets of kidney cancer is limited. In our previous work, we identified several splice variants (SVs) enriched in ccRCC.⁷ One of these SVs (chromosome 6, 167369679:167370715) was in the gene Ribonuclease T2 (RNASET2-SV) and was significantly associated with worse clinical prognosis across multiple patient cohorts. This finding inspired us to further study the biological and pathological implications of this alteration in ccRCC. RNASET2 is a widely conserved protein across species and is known to have independent functions both intracellularly and extracellularly.⁸ Intracellularly, it degrades and recycles cytoplasmic RNAs that are delivered to the lysosome/vacuole during autophagy.⁹ Extracellularly, it has been shown to serve as an alarmin molecule stimulating a local immune response to infection, inflammation, and cancer.^10–12 Few studies have reported on the role of RNASET2 in cancer biology. The majority of these studies have found it correlated with slower tumor growth.^11,13,14 Ovarian cancer is the most widely studied cancer model, with initial reports demonstrating decreased RNASET2 expression in more advanced tumors. Furthermore, in ovarian cancer, RNASET2 can exert a tumoricidal effect by modulating the extracellular tumor microenvironment, inducing immunocompetent macrophages to recognize and attack cancer cells.¹⁵ The impacts of RNASET2 have only recently been explored in ccRCC. In the only published study, the authors demonstrated that RNASET2 protein interacted with lipogenic pathways to promote tumorigenesis.¹⁶ Our study sought to better understand the spatial relationship of RNASET2 in ccRCC and the interplay with its splicing and the tumor immune microenvironment.

Materials and Methods

The Cancer Genome Atlas and Clinical Proteomic Tumor Analysis Consortium Cohorts

The normalized and debatched RNA-seq data from the The Cancer Genome Atlas (TCGA) PanCanAtlas publications were downloaded from https://gdc.cancer.gov/about-data/publications/ancanatlas using the file: “EBPlusPlusAdjustPANCAN_IlluminaHi-Seq_RNASeqV2.geneExp.tsv.” All expression values were log₂-transformed before analysis. The ccRCC samples were extracted based on publication by Ricketts et al.¹⁷ The TCGA survival data from Liu et al¹⁸ were used for survival analysis.

The University of ALabama at Birmingham CANcer data analysis Portal (http://ualcan.path.uab.edu/index.html) was used as an interactive web resource to aid in analyzing TCGA database.¹⁹

Secondary Analysis of Clinical Proteomic Tumor Analysis Consortium Clear Cell Renal Cell Carcinoma Cohort

We downloaded protein expression data for 103 ccRCC tumors in Clinical Proteomic Tumor Analysis Consortium (CPTAC) using the Proteomics Data Commons (https://pdc.cancer.gov/pdc/). We added the RNASET-SV-predicted amino acid sequence to a reviewed, human FASTA file from UniProt (www.uniprot.org) and processed the ccRCC CPTAC data using MaxQuant quantification software (version 1.6.14.0).²⁰ This was used to identify peptides using the UniProt human database (May 2022) and quantify the tandem mass tag reporter ion intensities. Up to 2 missed trypsin cleavages were allowed. The mass tolerance was 20 ppm in the first search and 4.5 ppm in the main search. Reporter ion mass tolerance was set to 0.003 Da. Minimal precursor intensity fraction was 0.75. Carbamidomethyl cysteine was set as fixed modification. Methionine oxidation was set as variable modification. Both peptide spectral match and protein false discovery rate were set at 0.01. Match between runs feature was activated to carry identifications across samples.

Total Cancer Care Cohort

From the years 2000 to 2018, primary tumor samples were obtained from patients with ccRCC who were surgically treated at Moffitt Cancer Center through protocols approved by the institutional review board (H. Lee Moffitt Cancer Center and Research Institute’s Total Cancer Care protocol [TCC] [MCC# 14690]; Advarra IRB Pro00014441). The diagnoses of ccRCC in our cohort were made by a genitourinary pathologist and confirmed by our study genitourinary pathologist (J.D.). Diagnosis was rendered when classic histological features of nested, tubular, or alveolar growth patterns with clear cytoplasm were present. Pertinent immunohistochemical stains such as CAIX, CD10, HMWCK, GATA3, AMACR, and TFE3 were utilized when there were morphological variations present and other renal cell tumors mimicking ccRCC would come in the differential diagnosis. The tumors were graded (1–4) using International Society of Urological Pathology/World Health Organization nuclear criteria.²¹ The American Joint Committee on Cancer, eighth edition, was used to evaluate tumor stage.²² Tumors underwent bulk RNA sequencing as part of this study. Clinicopathologic variables are tracked and maintained in an institutional database to perform analyses with genomic and immunohistochemical data. Written informed consent was obtained from all tissue donors. Bulk RNA sequencing of macrodissected tumor samples was performed using the TruSeq RNA Exome kit (Illumina) for 50 million 100-bp paired-end reads. RNA-seq reads were aligned to the human reference genome in a splice-aware fashion using Spliced Transcripts Alignment to a Reference, allowing for accurate alignments of sequences across introns.²³ Aligned sequences were assigned to exons using the HTSeq package to generate initial counts by region.²⁴ Normalization, expression modeling, and difference testing were performed using DESeq2.²⁵ Using Portcullis, we then screened the remaining SVs in our institutional TCC database, which includes 111 primary tumor samples from ccRCC patients.^7,26 SVs were excluded if they were detected in ≤5% of the total sample population. This cutoff was implemented to ensure that our selected SVs were well-represented in all cohorts; however, we acknowledge that it may exclude potentially relevant or interesting SVs that may have been underrepresented in certain cohorts. RNASET2-SV was identified by the RNA sequence at chromosome 6, 167369679:167370715.

Immunohistochemistry

Immunohistochemistry (IHC) staining for RNASET2, CD163 (marker for M2 macrophages), CD8 (marker for T cells), and TIM3 (marker for T cell exhaustion) was performed on tumor tissue from TCC patients.^27–29 Slides were deparaffinized on the Leica Bond RX automated system using dewax solution (Leica). Heat-induced antigen retrieval method was used in ER2 solution for 10 minutes (Leica).

For the CD163/RNASET2 co-stain, slides were stained using a Ventana Discovery XT automated system (Ventana Medical Systems) as per manufacturer’s protocol with proprietary reagents. Briefly, slides were deparaffinized on the automated system with EZ-Prep solution (Ventana). Heat-induced antigen retrieval method was used in Cell Conditioning 1 (Ventana). The rabbit primary antibody that reacts to RNASET2, (#LS-B8807, Lifespan Bio) was used at a 1:1000 titration and incubated for 32 minutes. The Ventana UltraMap Anti-Rabbit Secondary Antibody was used for 4 minutes. The detection system used was the Ventana ChromoMap Redkit. The slides were then sequentially stained with the rabbit antibody reactive to CD163 (#93498, Cell Signaling Technology) at a dilution of 1:500 and incubated for 32 minutes, followed by OmniMap anti-Rabbit multimer (Ventana) for 16 minutes and detection with ChromoMap DAB kit. Slides were then counterstained with hematoxylin. Slides were dehydrated and coverslipped as per normal laboratory protocol. The control for RNASET2 was kidney tissue (Supplementary Fig. S1). The control tissue for CD163 was non–small cell lung cancer. For the negative control of all markers, the antibody was omitted from being tittered.

For the CD8/TIM3 costain, slides were stained using a Ventana Discovery XT automated system (Ventana Medical Systems) as per manufacturer’s protocol with proprietary reagents. Briefly, slides were deparaffinized on the automated system with EZ-Prep solution (Ventana). Heat-induced antigen retrieval method was used in Cell Conditioning 1 (Ventana). The rabbit primary antibody that reacts to CD8 (#790–4460, Ventana) was used at a prediluted concentration and incubated for 8 minutes. The Ventana UltraMap Anti-Rabbit Secondary Antibody was used for 4 minutes. The detection system used was the Ventana ChromoMap Redkit. The slides were then sequentially stained with the rabbit antibody reactive to TIM3 (#45208, Cell Signaling Technology) at a dilution of 1:100 and incubated for 92 minutes, followed by OmniMap anti-Rabbit multimer (Ventana) for 16 minutes and detection with ChromoMap DAB kit. Slides were then counterstained with hematoxylin. Slides were dehydrated and coverslipped as per normal laboratory protocol. The control tissue for both was tonsil tissue. For the negative control of all markers, the antibody was omitted from being tittered.

Slides were digitized with a Leica Aperio AT2 slide scanner (Vista) using a 20 ×/0.75 NA objective lens. An experienced genitourinary pathologist (J.D.) used the annotation pen tool in Aperio ImageScope software to define the tumor-core zone in each hematoxylin and eosin image. High-resolution Aperio SVS images representing the 2 double-stained IHC slide sets (CD8/TIM3 and CD163/RNASET2) were imported into Visiopharm (Hoersholm) for quantitative digital image analysis. We leveraged a tissue alignment algorithm available in Visiopharm to optimally align the 2 adjacent slides from each patient sample set. These slides were visually inspected to ensure appropriate alignment. Within each aligned image set, 5 equally sized regions of interest (ROIs) were selected from the tumor edge and adjacent stroma. An additional 5 ROIs of the same size were selected from the tumor-core zone. The ROI selection process was performed to include regions that had a cellular appearance that was consistent with remainder of the slide and to avoid tissue artifacts. The ROI sizes were standardized at 900 × 900 pixels, at a pixel resolution of 0.502 μm/pixel. Thresholds for staining positivity were set by an experienced digital pathology image analyst (J.J.) and confirmed by a study pathologist (J.D.). Thresholding on the DAB (AEC-DAB) and red (triple red) color channels (inverted so that stronger staining equals higher intensity) was used to label cells as negative, DAB positive, and red positive. For the DAB channel, intensity values in the brightest 20% of pixels were taken—if between 0 and 75, then the stain was negative, and if between 75 and 255, then the stain was considered positive. For the red channel, intensity values in the brightest 50% of pixels were taken—if between 0 and 50, then the stain was negative, and if between 50 and 255, then the stain was considered positive. In the case of dual staining, DAB-labeled objects met criteria for the red threshold mentioned above, and red-labeled objects must have an intensity value between 50 and 255 in brightest 5% of pixels that contain DAB.

For H score thresholding, all red and dual objects were reclassified based on the triple-red intensity in the brightest 20% of pixels within the cell object. Intensity values between 50 and 115 were considered weak. Intensity values between 115 and 180 were considered moderate. Intensity values between 180 and 255 were considered strong. All thresholds are identical for each image in the study. Specific thresholds were chosen under the guidance of Dr. Dhillon (her impression of general staining on the whole slide) and tested by image analysis specialists. This included arbitrarily choosing thresholds and adjusting them based on visual performance when comparing across samples. The technician was blinded to patient data and outcomes and was not biased to specific samples. Choosing 20% or 50% of the pixels was due to distribution of stains within the cell body. The RNASET2 tends to be spots that are the brightest pixels in the cell. On the other hand, the cell growth algorithm does not always capture the accurate shape of a macrophage so as long as there is some bright DAB staining in part of the cell, it will be considered positive.

These thresholds were used with Visiopharm’s cell detection algorithms to identify and categorize cells as positive for each biomarker, including coexpression, or negative on the basis of staining intensity. Spatial coordinates and mean fluorescence intensity values were extracted for each cell, nucleus, cytoplasm, and extracellular area. Additional classification of TIM3- and RNASET2-positive cells was performed to segment the cells into 4 intensity categories: negative, weak, moderate, and strong. This intensity-based distribution of cells is comparable with the qualitative method used by pathologists, in which staining intensity is categorized as 0, 1+, 2+, and 3+ (24). Percent positivity and H scores were calculated for each ROI using these intensity data. H score is an approach that globally quantifies intensity and percent positivity throughout the entire ROI into 1 score, according to the following formula: H score = (1 × (%cells weak)) + (2 × (%cells moderate)) + (3 × (%cells strong)) (25). H scores range from 0 to 300, with 0 representing no cell staining for the marker of interest and 300 representing every cell staining with the highest intensity (26). In addition, the general workflow of our method can be seen in Supplementary Figure S2.

Cell Line Experiments

Human ccRCC cell lines including 786-O and A498 were used for in vitro experiments. These cell lines were obtained from Dr. Timothy Shaw lab at the Moffitt Cancer Center. Genomic profiling of these cell lines has proven them to be most consistent with the clear cell subtype.^30,31 These cell lines selected after analysis of the Cancer Cell Lines Encyclopedia database demonstrated high expression of RNASET2 and the corresponding presence of RNASET2-SV in 786-O and A498. A lentiviral vector using 2 shRNAs was used to perform RNASET2 knockdown experiments and subsequently confirmed with Quantitative PCR (qPCR). 786-O was cultured in RPMI-1640 medium (Gibco) with 10% fetal bovine serum. A498 cells were cultured in Dulbecco’s Modified Eagle Medium medium (Gibco) containing 10% fetal bovine serum. The culture medium was supplemented with 100 U/mL penicillin and 100 mg/mL streptomycin to avoid bacterial contamination. All cells were cultured in humidified air at 35 °C with 5% CO₂.

We designed 2 shRNA clones to knock down RNASET2 gene expression utilizing lentiviral vectors from VectorBuilder Inc. The shRNASET2 clone1 sequence was: 5′-CACCACAGTGGAAGCCAGGTGCCTTTAATCCACTGTAACC-3′. The shRNASET2 clone2 sequence was: 5′-CAGCAATGCACCAATGCAATGCAGGTCATTAACAATTATCA-3′. The shRNA scramble control sequence was: 5′-CCTAAGGTTAAGTCGCCCTCGCTCGAGCGAGGGCGACTTAACCTTAGG-3′. Lentiviral particles were packaged in the 293T cells as previously described.³² Cells were harvested after puromycin selection for 3 days for qPCR and other experiments.

To transiently overexpress RNASET2, an mCherry/Neo-tagged pRP mammalian gene expression vector encoding wild-type RNASET2 was transfected into 786-O and A498 cell lines. More specifically, cells were seeded in 6-well plates and transfected with 2.5 μg of vector using Lipofectamine 3000. The vector used, pRP[Exp]-mCherry/Neo-CAG>hRNASET2[NM_003730.6], was constructed and packaged by VectorBuilder. The vector ID is VB230531–1272kvw, which can be used to retrieve detailed information about the vector on vectorbuilder.com. Following a 48-hour incubation, cells were subjected to IncuCyte analysis with 5000 cells per well. Total protein extraction was conducted and electrophoresis was performed as described previously using Tricine Sample Buffer and Mini-PROTEAN Tris/Glycine Precast Gels (BioRad), with minor adjustments.³¹ Luminescent signals were generated using SuperSignal West Pico Plus Chemiluminescent Substrate (Thermo Fisher Scientific) and visualized using a LI-COR imaging system.

Cell Proliferation Assay

The effect of RNASET2 knockdown on cell proliferation was determined using an Incucyte ZOOM system. The Incucyte ZOOM system (Essen Bioscience) allowed the monitoring of the proliferation of both control and treated/experimental groups in a defined environment of a standard incubator. All experimental groups were seeded in a 96-well plate with 6 replicates per group at a concentration of 1000 cells/well and monitored for 5 days.

RNA Sequencing and Downstream Analyses

To determine the effect of RNASET2 on signaling pathways, we transfected A498 and 786-O cell lines as described above. Bulk RNA sequencing was performed for both control and RNASET2 knockdown cells. Total RNA was isolated using total RNA extraction kit (Zymo Research). The RNA concentration was determined by measuring the absorbance at 260 nm using a NanoDrop (Thermo Scientific). High-quality RNA samples were sent to Novogene (Chula Vista) for RNA library preparation and sequencing. 150 bp pair-end sequencing was performed in a HiSeq2500 (Illumina). The processed reads were aligned against the human genome assembly hg38 using RNA-seq by Expectation-Maximization with default settings.³³ Differential expression analysis was performed from expected read count matrix using EBSeq.²⁵ We applied gene set enrichment analysis (GSEA) to interpret the RNA-seq results upon knockdown of RNASET2 in gene set collections.³⁴ Online Database for Annotation, Visualization, and Integrated Discovery functional microarray analysis tool (https://david.ncifcrf.gov/summary.jsp) was used for pathway enrichment analysis. False discovery rate adjustment was applied for multiple testing corrections.

Statistical Analyses

Statistical analyses were performed using R program version 4.1.1 (R-project). The Wilcoxon (2 groups) or Kruskal–Wallis (3 or more groups) tests were used to compare RNASET2 H score against different clinical/demographic variables among the different ROIs; for continuous demographic variables like age, a Pearson correlation test was computed. The Kaplan-Meier curve method was used to estimate a survival distribution, and the logrank test was performed to compare survival distributions after stratifying patients into groups of high- or low-density levels based on the median of individual markers as a cutoff or several clinical/demographic variables. Univariate Cox proportional hazards regression analyses were also performed to compute the hazard ratio of quantitative marker densities for risk of death while controlling for age, race, sex, pathological stage (American Joint Committee on Cancer staging system), and histological grade.

Results

Transcriptomic and Proteomic RNASET2 Expression in The Cancer Genome Atlas and Clinical Proteomic Tumor Analysis Consortium Cohorts

We first chose to define the relationship between wild-type RNASET2 expression and ccRCC tumorigenesis. The Kidney Renal Clear Cell Carcinoma cohort from the TCGA database was queried. When comparing normal renal tissue with tumor tissue (Fig. 1A), we found that tumors had significantly higher gene expression of RNASET2 (P < .001). Similarly, when comparing histologic grade, increasing RNASET2 expression was positively correlated with increasing grade (Fig. 1B). The cohort was then stratified into those patients with low RNASET2 expression (normalized expression <12.3) vs high RNASET2 (normalized expression >12.3) expression utilizing maximally selected rank statistics. A Kaplan-Meier analysis was then performed. We found that there was a significant difference in overall survival between the 2 groups, with high RNASET2 expression associated with worse survival (P < .001) (Fig. 1C).

Figure 1.

RNASET2 expression in The Cancer Genome Atlas and Clinical Proteomic Tumor Analysis Consortium cohorts. (A) RNA expression of RNASET2 in primary tumor compared with normal renal tissue in The Cancer Genome Atlas database; *P < .001. (B) RNA Expression of RNASET2 stratified by tumor grade; ****P < .05 for grade 4 vs grade 2, grade 4 vs grade 1, and grade 4 vs normal; ***P < .05 for grade 3 vs normal; **P < .05 for grade 2 vs normal; *P < .05 for grade 1 vs normal. (C) Kaplan-Meier curve demonstrating the overall survival difference between high RNASET2-expressing tumors and low RNASET2-expressing tumors; P < .001. (D) Correlation of RNASET2 gene expression with expression of RNASET2-SV using The Cancer Genome Atlas cohort; P = 1.21 × 10⁻²³.

We examined the impact of the RNASET2-SV on transcriptomic changes in kidney renal clear cell carcinoma and found a moderate positive correlation with increasing RNASET2 wild type gene expression with increasing RNASET2 SV (Spearmen r = 0.414; P = 1.21×10⁻²³; Fig. 1D)

Proteomic and Epigenetic Changes Resulting From RNASET2-SV

Based on the spliced RNA sequence at chromosome 6, 167369679:167370715, we predicted the resulting amino acid sequence. The splice junction predicted an out-of-frame peptide truncated 63 amino acids after the splice junction amino acid (Supplementary Fig. S3). We first queried Protein Basic Local Alignment Search Tool (https://blast.ncbi.nlm.nih.gov/) for similar protein sequences in humans, but there were no matches. Next, we searched the CPTAC database for the peptide but did not detect evidence of a novel RNASET-SV protein, suggesting that RNASET2-SV is not translated into a protein.

Given the strong association of increased RNASET2 protein with the presence of the RNASET2-SV, we specifically examined the location of RNASET2-SV in the context of epigenetic changes to determine if splicing might affect the chromatin structure that would lead to increased transcriptional efficiency of the wild-type protein. Figure 2 depicts the location of the splicing event in the 786-O ccRCC cell line, as well as the surrounding methylation sites and histone marks (Fig. 2A, C). Based on corroborating data from CPTAC and TCGA, the splicing event is located within hypermethylated regions, which, once spliced out, create a hypomethylated state (Fig. 2B). Furthermore, upstream of the splicing event are acetylation sites of H327K (H3K27ac), a well-recognized epigenetic marker for enhancing regions (Fig. 2D). Bringing the gene in closer proximity to this enhancer region may lead to its increased expression.

Figure 2.

RNASET2 location and surrounding epigenetic changes. This figure represents the location of RNASET2 in the 786-O clear cell renal cell carcinoma cell line. (A) The first exon of RNASET2 is represented by the green bar, with the remaining exons transcribed in the direction of the arrows. The location of the RNASET2 splicing junction is represented by the pink circle, upstream of the first exon. (B) Methylation sites from Clinical Proteomic Tumor Analysis Consortium (CPTAC) and The Cancer Genome Atlas (TCGA) cohorts that are located within the RNASET2 gene are overlayed. Blue lines indicate a hypermethylated site. Red lines indicate a hypomethylated site. The boxed region, consisting mainly of hypermethylation, is spliced out in RNASET2-SV. Surrounding this region are multiple areas of hypomethylation that remain within the transcribed gene, increasing transcription. (C) Sequencing of the 786-O clear cell renal cell carcinoma cell line validates RNASET2-SV by identifying multiple reads of the 1036 base-pair splicing event. Validation of the splicing event in the 786-O cell line. (D) These blue peaks represent sites of enriched acetylated histone 3 (H3K27ac), epigenetic marks utilized to identify enhancer regions. The higher peaks are indicative greater enrichment and thus upregulated transcription.

Immunohistochemistry Analysis Reveals That RNASET2-SV Is Associated With Increased RNASET2 Protein Expression

We then examined the relationship of RNASET2 within the local tumor immune cell populations (Fig. 3). Using IHC staining of ccRCC tumors from our institutional cohort, we measured RNASET2 protein expression in addition to the expression of relevant immune cell markers, including CD163 (M2 macrophages), CD8 (T cells), and TIM3 (exhausted T cells). Given the tumor heterogeneity of ccRCC, we examined changes in staining from 5 ROIs across 3 different microscopic sections of the tumor including the tumor core, tumor-stroma interface (tumor edge), and stroma, for a total of 15 ROIs per patient. This cohort had been previously characterized for the presence of the RNASET2-SV using bulk RNA sequencing. Among the 75 patients analyzed, the RNASET2-SV was present in 60 patients using a cutoff value of greater than 1% of reads mapping to the SV neojunction. The clinical characteristics of this cohort can be found in Table 1, consisting mainly of patients with high-risk, locally advanced diseases.

Figure 3.

RNASET2/CD163 co-staining across stages and regions of interest. As there are increases in stage, there is an increase in protein expression of RNASET2. Spatially, RNASET2 expression increases from stroma and tumor edge to the tumor core.

Table 1.

Baseline patient and specimen demographics

Characteristic	Total (n = 75)
Age, y	65.8
Median interquartile range	57.1–72.0
Sex—no. (%)
Male	51 (68.0)
Female	24 (32.0)
Race—no. (%)
White	66 (88.0)
Non-White	9 (12.0)
Tumor size (cm)
Median	6.5
Interquartile range	4.75–9.10
Grade—no. (%)
Low grade (1–2)	18 (24.0)
High grade (3–4)	57 (74.0)
Stage—no. (%)
I/II	17 (22.7)
III	40 (53.3)
IV	18 (24.0)
RNASET2-SV
Positive	60 (80)
Negative	15 (20)
Death
Yes	21 (28.0)
No	54 (72.0)
Recurrence—no. (%)
Yes	18 (24.0)
No	57 (74.0)

We found that across the entire cohort, there was a positive correlation between the wild-type RNASET2 RNA expression and expression of the RNASET2-SV (R = 0.46; P < .001). High RNASET2 was associated with male sex in the tumor core (P < .001) and tumor edge (P < .001). RNASET2 protein expression was significantly higher in grade 4 tumors in the stroma (P = .0055) and tumor edge (P = .024), but not in tumor core (P = .12). Across all tumor regions, RNASET2 did not correlate with patient age, race, or tumor stage (Supplementary Fig. S4).

We then correlated expression levels of the wild-type RNASET2 protein with the presence/absence of the RNASET2-SV. Tumors with RNASET2-SV had significantly higher levels of RNASET2 protein expression, relative to SV-negative tumors in the tumor core (P = 3.3 × 10⁻⁵) and tumor edge (P = 3.8 × 10⁻³; Fig. 4A, B). If only RNASET2-SV-negative tumors with high wild-type RNA expression (defined as being greater than the median of all RNASET2-SV-negative tumors) are compared with all RNASET2-SV-positive tumors, the presence of the RNASET2-SV still leads to higher protein expression in the tumor core (P = .0084). These results were validated when examining the CPTAC data, supporting increased RNASET2 protein expression in SV-positive tumors (P = 2.95 × 10⁻¹²; Fig. 4D).

Figure 4.

RNASET2-SV is associated with higher RNASET2 protein expression. (A) Immunohistochemistry slides containing RNASET2 and CD163 staining from 2 patients, both of whom had stage IV disease. The image on the left represents tumor tissue obtained from the nephrectomy specimen of a 57-year-old man with low RNASET2 expression and associated prolonged survival of 9 years. The image on the right represents tumor tissue obtained from the nephrectomy specimen of a 65-year-old man with high RNASET2 expression and associated with poor overall survival of only 1 year. (B) RNASET2 protein expression stratified by region of interest (stroma, tumor edge, and tumor core) and presence/absence of RNASET2-SV. (C) Intracellular:extracellular RNASET2 protein expression stratified by region of interest (stroma, tumor edge, and tumor core) and the presence/absence of RNASET2-SV. (D) RNASET2 protein expression stratified by presence/absence of RNASET2-SV based on Clinical Proteomic Tumor Analysis Consortium cohort; P = 2.95 × 10⁻¹².

RNASET2 Localizes to the Intracellular Compartment in Splice Variant-Positive Tumors

We then examined the subcellular localization of RNASET2 by differentiating extracellular and intracellular expressions. When comparing RNASET2-SV-positive and -negative tumors, we found that positive tumors had a significantly higher ratio of intracellular to extracellular RNASET2 expression in both the tumor core (P = .0062) and tumor edge (P = .00012; Fig. 4C).

RNASET2 and Interaction With the Tumor Immune Microenvironment

There was a significant inverse correlation between RNASET2 and CD163 in the tumor edge (R = −0.3; P = .0067) and tumor core (R = −0.39; P = .00026), but not in the stroma (R = −0.06; P = .59; Supplementary Fig. S5A). When this is stratified based on the presence of the RNASET2-SV, only the tumor core of SV-positive patients retains this significant association between RNASET2 and CD163 (R = −0.41; P = .0013). There is a significant correlation between RNASET2 and TIM3 in the stroma (R = 0.22; P = .045), but not in tumor edge (R = 0.077; P = .49) or tumor core (R = −0.029; P = .79; Supplementary Fig. S5B). When this is stratified by the presence/absence of the SV, there was no change in this relationship. There was no direct relationship between the presence of RNASET2-SV and expression of TIM3, CD163, or CD8 across all tumor regions. In addition, by using the xCell gene signature-based method, we evaluated the relationship of the SV with immune cell types (Supplementary Table S1). We again found no direct correlation between the RNASET2-SV and the estimated immune populations.

Cox Proportional Hazards Model

A Cox proportional hazards model was then developed to explore the impact of age, race, sex, tumor grade, tumor stage, and RNASET2 expression (Table 2). On univariate analysis, only increasing grades proved to predict worse overall survival. However, there was a signal to suggest that higher intracellular RNASET2 expression within the tumor edge and tumor core predicted worse overall survival (tumor edge: HR = 10.97; 95% CI, 0.92, 130.75; P = .058; tumor core: HR = 10.25; 95% CI, 0.881, 119.21; P = .063).

Table 2.

Univariate Cox proportional hazards model for overall survival (n = 75)

Univariate cox proportional hazards regression models for overall survival
Variable	Level	HR (95% CI)	P value	n (events)
Age (y)		1.006 (0.965, 1.048)	.7884	75 (21)
Race	Other	1.0 (Reference)		75 (21)
	White	2.491 (0.332, 18.694)	.3747
Sex	Female	1.0 (Reference)		75 (21)
	Male	2.387 (0.795, 7.169)	.1210
Pathologic stage	I/II	1.0 (Reference)		75 (21)
	III	1.302 (0.274, 6.176)	.7397
	IV	3.582 (0.774, 16.576)	.1026
Histologic grade	2	1.0 (Reference)		75 (21)
	3	3.960 (0.884, 17.727)	.0719
	4	8.472 (1.338, 53.645)	.0233
RNASET2 expression (stroma)		0.975 (0.698, 1.363)	.8840	75 (21)
RNASET2 expression (tumor edge)		1.402 (0.966, 2.033)	.0753	75 (21)
RNASET2 expression (tumor core)		1.423 (0.948, 2.135)	.0884	75 (21)
RNASET2 I:E expression (stroma)		2.809 (0.042, 189.199)	.6306	74 (21)
RNASET2 I:E expression (tumor edge)		10.966 (0.920, 130.748)	.0583	75 (21)
RNASET2 I:E expression (tumor core)		10.247 (0.881, 119.208)	.0631	75 (21)

HR is odds of death.

I:E represents the intracellular to extracellular ratio of RNASET2.

RNASET2 Promotes Clear Cell Renal Cell Carcinoma Cell Line Proliferation Through Altered Metabolic Pathways

We performed knockdown experiments in A498 and 786-O (Fig. 5A, Supplementary Table S2). RNASET2 knockdown significantly inhibited A498 and 786-O cell proliferation (Fig. 5B, C). RNASET2 was subsequently overexpressed in these cell lines, demonstrating a proliferative trend in both cells using the RNASET2 plasmid compared with control plasmid-transfected cells (Supplementary Fig. S6). Interestingly, this trend was slightly more robust in A-498 cells, possibly due to the notably higher baseline RNASET2 expression in the 786-O cell line.

Figure 5.

RNASET2 knockdown decreases cell proliferation in clear cell renal cell carcinoma cell lines. (A) Quantitative PCR and Western blot demonstrating successful knockdown of RNASET2 in A498 and 786-O using 2 shRNAs. Control and Knockdown cell lines for 786-O (B) and A498 (C) were seeded at a density of 1 × 10³ cells/well in 96-well plates with 6 replicates. At the end of 5 days, cell proliferation rates were analyzed using Incucyte software.

We performed RNA-seq analysis of differentially expressed genes upon depletion of RNASET2 in A498 and 786-O. To investigate potential functional categories of RNASET2 knockdown target genes, we performed GSEA and Gene Ontology and revealed several pathways highly enriched in upregulated genes after RNASET2 knockdown. In GSEA, both the cholesterol biosynthesis pathway (normalized enrichment score, 1.64; P = .001) and Kyoto Encyclopedia of Genes and Genomes Steroid biosynthesis pathway (normalized enrichment score = 1.61; P < .001) were upregulated in the control compared with knockdown (Fig. 6A). The top 3 ontologies in A498 were membrane, plasma membrane, and lipid metabolism. In 786-O, the top 3 ontologies were glycoprotein, signal, and CARBOHYD: N-linked (GlcNAc.) asparagine. However, it did share common ontologies with A498, including membrane, cell membrane, and plasma membrane (Fig. 6).

Figure 6.

Gene set enrichment analysis and Gene Ontology of signaling pathways activated during ccRCC. RNASET2 expression had enrichment of the (A) cholesterol biosynthesis pathway and (B) Kyoto Encyclopedia of Genes and Genomes (KEGG)—Steroid Biosynthesis. Heat-map diagrams are utilized to display significant gene set enrichment analysis pathways. Color gradient represent expression level (red = expression and blue = lower expression). (C) Gene Ontology enrichment analysis showing top 20 pathways associated with RNASET2 expression with respect to cellular component, biologic pathways, and molecular function.

Discussion

In our study, we identified RNASET2 to be a biologically active and spatially diverse protein that could be contributing to tumor progression in a subset of ccRCC patients. In addition, we found evidence to suggest that RNASET2-SV is not translated into a functional protein but correlates with higher wild-type RNASET2 protein expression, possibly through epigenetic modifications. Our RNA sequencing and gene pathway enrichment analysis suggests that this pathogenic contribution is mainly through lipid metabolism, based on changes in gene expression.

An extensive search of the predicted peptide sequence for the RNASET2-SV did not reveal it to be a translated protein. However, given the novelty of this peptide, it is possible that the shotgun proteomics approach utilized in these databases was not calibrated to identify it. The tandem mass tag approach has a strong selection bias for proteins (and their peptides) that are expressed in every sample in the batch; proteins like the RNASET2-SV may not be observed in these experiments. Based on its location, as depicted in Figure 2, RNASET2-SV is found within an enhancer region, which regulates gene expression, consistent with the lack of evidence of RNASET2-SV detection at the protein level. Additionally, within this enhancer region upstream to RNASET2, there is an enrichment of H3K27Ac, histone marks that can be modified to unpack the chromatic structure and increase gene expression. Indeed, in our previous publication, multiple CpG probes for RNASET2 gene were increasingly hypomethylated as the splice fraction for RNASET2-SV increased.⁷ This result raises the possibility that RNASET2-SV may act as an enhancer RNA. These enhancer RNAs are typically described as short, unstable transcripts that remain in the nucleus and act as transcription activators to promote the expression of a target gene, in this case, RNASET2.³⁵ Further work is required to directly implicate that RNASET2-SV is interacting with the chromatin machinery involved in RNASET2 expression.

Given the association observed in prior studies on solid tumors that RNASET2 influences the tumor immune microenvironment, especially macrophages, we chose to perform IHC on ccRCC tumor samples. This result in turn allowed us to examine expression information in conjunction with patient-level SV data. As was predicted from the transcript level data, RNASET2-SV was correlated with increased wild type RNASET2 protein production. Furthermore, the protein preferentially localizes within the cell and has a significant association with the immune cell marker CD163. Despite the positive correlation with M2 macrophages, the overall association between RNASET2 and immune cell markers was inconsistent. In contrast to other studies in solid tumors, these results support point toward the role of RNASET2 in ccRCC to drive oncogenic pathways intracellularly, with a lesser role in interacting with the extracellular immune cell populations.^11,13

We subsequently confirmed the oncogenic effect with RNASET2 knockdown in A498 and 786-O cell lines, which is also supported by findings reported in a study by Quan et al.¹⁶ RNA sequencing and GSEA revealed that cholesterol biosynthesis pathways are enriched in the knockdown cell lines compared with their controls. This expands upon the findings of Quan et al, in which the authors found that RNASET2 regulates enzymes critical to triglyceride synthesis in ccRCC cell lines. Furthermore, using Gene Ontology, we identified pathways involved in lipid metabolism and plasma membrane functions in A498 and 786-O, corroborating the findings from GSEA. The results of RNA sequencing suggest that through upregulation of these lipid pathways, these cells might be utilizing alternative sources of energy for tumor growth and proliferation. In summary, we demonstrated that RNASET2 exhibits a tumor-promoting role in the pathogenesis of ccRCC. Furthermore, we have provided a rationale for how the RNASET2-SV leads to an upregulation of wild-type RNASET2 protein expression and that this upregulation leads to alterations in lipid metabolism that drive cell proliferation.

Supplementary Material

The online version contains supplementary material available at https://doi.org/10.1016/j.labinv.2024.102041

Data Availability

The data used to support the findings of this study are available from the corresponding author upon reasonable request.